Informatics Help Desk
Weekly on Wed/Thu 9:30am to 11:00am (central time);
Fri 1:00pm to 2:30pm (central time)
[Remote via Zoom]
Passcode: 541538
What is This?
We are a NSF supported service to help solve technical issues and streamline research in materials and chemistry informatics. We have experts in data science application to materials and chemistry informatics who would love to help answer any questions you have. No question is too basic and we will do our best with really hard questions. We are open to everyone!
Hands-On Tutorials
In addition to the regular drop-in style Zoom calls, we are also offering regular short targeted hands-on tutorials with the aim of getting researchers up to speed with state-of-the-art software, techniques, and ideas . Each tutorial will aim to have a short technical introduction and an interactive exploration. These tutorials are typically with small groups and offer a chance to ask detailed questions and make a major step in one's knowledge and skills.
Recordings from each tutorial will be uploaded for those that aren't able to make them in person.
Previous Tutorials
March 30, 2023 - Introduction to GPU Acceleration for Python
Speaker: Simon Delattre
Who: Beginner ML researchers with some Python background
What: An introductory example to using the Jax python package to easily utilize GPU resources with existing python code. Simon will take us through an example that can hopefully be modified to work with any code attendees are working on.
When: March 30 at 10:00 am central time.
Where: In the normal help-desk Zoom call. We'll create a breakout room for workshop specific attendees to separate out from general drop-in questions.
Links: PowerPoint Slides
March 15, 2023 - Introduction to Deep Learning With Keras
Speaker: Professor Dane Morgan
Who: New or prospective machine learning researchers
What: Keras is a popular package for building deep learning neural networks. In this introduction we'll show an example of how to get started with this python package and talk about some common best practices.
When: March 15 at 10:00 am central time.
Where: In the normal help-desk Zoom call. We'll create a breakout room for workshop specific attendees to separate out from general drop-in questions.
Links: Day 2 from the Summer boot camp materials
March 1, 2023 - Estimating Errors in Machine Learning Models
Speaker: Dr. Ryan Jacobs
Who: Current ML Researchers
What: Obtaining accurate error bars on a model's predictive performance is necessary to establishing a model's performance. We'll introduce a technique for establishing and assessing the quality of a model's error bars to more confidently make new predictions.
When: March 1 10:00 am central time.
Where: In the normal help-desk Zoom call. We'll create a breakout room for workshop specific attendees to separate out from general drop-in questions.
Links: Tutorial 6 from the MAST-ML tutorials here
February 15, 2023 - Introduction to Machine Learning - round 2
Speaker: Dr. Benjamin Afflerbach
Who: New or prospective machine learning researchers
What: An introduction to common steps in a research workflow to start with a new dataset and end up with a trained and assessed machine learning model. We'll introduce ideas for data cleaning, featurization, a common first model to try (random forests), model assessment, optimization, and making predictions. The activity will use scikit-learn as the primary ML software package, and is entirely cloud based so you do not need to do any prep to install anything before or during the activity. If you'd like to preview the content or work through independently you can find the resource hosted on Nanohub at the link below.
When: February 15 at 10:00 am central time.
Where: In the normal help-desk Zoom call. We'll create a breakout room for workshop specific attendees to separate out from general drop-in questions.
Links: https://nanohub.org/tools/intromllab
Tutorial Recording: https://youtu.be/jO0E9XEH34k
Nov. 30, 2022 - Introduction to Machine Learning
Speaker: Dr. Benjamin Afflerbach
Who: New or prospective machine learning researchers
What: An introduction to common steps in a research workflow to start with a new dataset and end up with a trained and assessed machine learning model. We'll introduce ideas for data cleaning, featurization, a common first model to try (random forests), model assessment, optimization, and making predictions. The activity will use scikit-learn as the primary ML software package, and is entirely cloud based so you do not need to do any prep to install anything before or during the activity. If you'd like to preview the content or work through independently you can find the resource hosted on Nanohub at the link below.
When: Nov. 30th at 9:30 am central time. The planned introductory presentation will take ~15 minutes.
Where: In the normal help-desk Zoom call. We'll create a breakout room for workshop specific attendees to separate out from general drop-in questions.
Help Desk Volunteers
Benjamin Afflerbach, Phd
Ben is a postdoc in the Computational Materials Science group at UW-Madison. His work includes using machine learning models to predict materials properties and he helps manage the undergraduate research group the Informatics Skunkworks
Simon Delattre, Phd
Simon is an engineer at Penn State University. He assists researchers across disciplines in leveraging deep learning and gaussian processes.
Logan Ward, Phd
Logan Ward is a staff scientist at Argonne National Laboratory's Data Science and Learning Division. Logan has a decade of experience in implementing materials informatics methods including both classic and deep-learning techniques. He is also an active developer on many open-source Python libraries for data management and scientific computing.
This resource supported by the National Science Foundation under Grant No. (2017072) and Grant No. (2020243). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.